The shift to electronic publishing of scientific journals is an opportunity for the digital library to provide non-traditional
ways of accessing the literature. One method is to use citation metadata drawn from a collection of electronic journals to
generate maps of science. These maps visualize the communication patterns in the collection, giving the user an easy-tograsp
view of the semantic structure underlying the scientific literature. For this visualization to be understandable the complexity of the citation network must be reduced through an algorithm. This paper describes the Citation Pathfinder application and its integration into a prototype digital library. This application generates small-scale citation networks that expand upon the search results of the digital library. These domain maps are linked to the collection, creating an interface that is based on the communication patterns in science. The Main Path Analysis technique is employed to
simplify these networks into linear, sequential structures. By identifying patterns that characterize the evolution of the
research field, Citation Pathfinder uses citations to give users a deeper understanding of the scientific literature.